Open wg21bot opened 1 month ago
Chair: Matthias Kretz
Champion: Damien Lebrun-Grandie
Minute Taker: Frank Birbacher
There was a concern that the name "nand" is confusing; and that the function is unnecessary if it implements an "and-not" - if anything, it needs to be a "not-and". The paper should add motivation for why a "not-and" is useful.
Also the paper should spell out how operations on floating-point resulting in NaN work. The expectation in SG6 was bit-wise comparison, rather than floating-point comparison (in a CAS loop).
SG6 sees no need to see the next revision of this paper. There is no real numerics concern.
No objection to more work on D3306R1 based on the minuted feedback provided
P3306R0 Atomic Read-Modify-Write Improvements (Gonzalo Brito Gadeschi, Damien Lebrun-Grandie)